I have yet to see a crash from such an issue, but it could clearly
happen, so let's make it safe

This avoids trying to use layers for which vbos are not ready yet

Normals are not defined for travel moves, so we need to enable the array
after drawing them, not before

This way we can easily yield at the end of each iteration. The remaining
issue is mostly how to handle the costly numpy.fromiter, which we don't
want to do uselessly several times.

But in my world, you have nothing to fear.